Atypical Unfolding of Early Brain Development in Autism: A Diffusion Tensor Imaging Study of Structural Connectivity and White Matter Organization

Saturday, May 17, 2014: 2:45 PM
Marquis BC (Marriott Marquis Atlanta)
R. Verma1, Y. Ghanbari*1, W. Parker2, M. Ingalhalikar2, M. A. Styner3, G. Gerig4, J. D. Lewis5, J. R. Pruett6, A. M. Estes7, A. C. Evans8, S. Das8, P. Kostopoulos8, H. C. Hazlett3, S. J. Paterson9, J. Pandey9, H. Gu10, K. N. Botteron11, S. Dager12, R. C. McKinstry6, R. T. Schultz9, J. Piven3 and .. The IBIS Network13, (1)*Joint first authors, University of Pennsylvania, Philadelphia, PA, (2)University of Pennsylvania, Philadelphia, PA, (3)University of North Carolina at Chapel Hill, Chapel Hill, NC, (4)School of Computing & Scientific Computing and Imaging Institute SCI, University of Utah, Salt Lake City, UT, (5)McGill University, Montreal, QC, Canada, (6)Washington University School of Medicine, Saint Louis, MO, (7)Speech and Hearing Sciences, University of Washington, Seattle, WA, (8)Montreal Neurological Institute, McGill University, Montreal, QC, Canada, (9)Center for Autism Research, The Children's Hospital of Philadelphia, Philadelphia, PA, (10)UNC Chapel Hill, Chapel Hill, NC, (11)Psychiatry and Radiology, Washington University School of Medicine, Saint Louis, MO, (12)University of Washington, Seattle, WA, (13)Autism Center of Excellence, Chapel Hill, NC

Investigating changes in brain connectivity and white matter (WM) integrity in infants who later develop autism spectrum disorder (ASD) should provide important insights into underlying neural mechanisms and early biomarkers at a time when autistic behaviors are first emerging. 


This study employed diffusion tensor imaging (DTI) to examine structural connectivity and measures of WM tissue organization like fractional anisotropy (FA) and mean diffusivity (MD), prospectively, in male infants at high familial risk for ASD (who did and did not meet clinical best-estimate (CBE) criteria for ASD at 24 months (HR+ and HR-, respectively) and non-familial low risk controls (LR-), who did not meet CBE criteria for ASD, at 6, 12 and 24 months of age.   


DTI data was acquired on 160 HR and 55 LR infants at 6, 12 and 24 months, including 42 HR+ and 118 HR- subjects classified with and without ASD, respectively, at 24 months. Images were registered to a population-specific atlas, subject-wise FA and MD images were created and a linear regression model was fitted per voxel with age, group, and age-by-group interaction incorporated and corrected for multiple comparisons. Automated full brain streamline tracking was constrained with a 314-region parcellation to create a 314x314 connectome, with inter-regional connectivity quantified by the number of connecting fibers, normalized to the template. Connection-wise t-tests on matrices were examined between groups at each age. 


Decreased connectivity was broad spread at 6 and 12 months in HR+ vs. LR-, becoming more specific to decreased long-range connectivity between frontal and posterior regions at 24 months. In concert with decreased long-range connectivity, increased short-range (within-hemisphere) connectivity was observed in HR+ vs. LR- subjects at 24 months. With respect to voxel level measures of WM integrity, HR+ (vs. LR-) subjects had increased MD in right auditory regions of Heschl’s gyrus, superior temporal gyrus and the acoustic radiation at 6 months; increased FA in primary somatosensory regions of right postcentral and supramarginal gyri at 12 months; and, increased MD in the left parahippocampal gyrus, hippocampus, supramarginal gyrus, superior temporal gyrus and acoustic radiation at 24 months.


Using voxel level and connectome analyses, we found patterns of abnormal brain development from 6 to 24 months in HR infants who met CBE criteria for ASD.  With voxel level analyses, HR+ showed delayed WM maturation in primary sensory-motor cortices in ASD (regions underlying early information processing) followed by later disorganization of higher order language, learning and memory regions at 24 months. Connectivity analyses showed wide spread decreased connectivity in HR+ at 6 and 12 months of age, but more selective connectivity deficits at 24 months, affecting only long-range connections.  Interestingly, at 24 months HR+ showed greater short-range connectivity. However, HR- did not show decreased long or short range-connectivity at any age, suggesting that decreased long-range connectivity by 24 months is critically important to the development of ASD in at-risk infants.  Ongoing analyses are looking at relationships between voxel level and connectome measures to improve our understanding of causal mechanisms and develop predictive biomarkers.